One of the greatest challenges IT organizations face going into 2011 is the burgeoning of data that, worldwide, should eclipse 1,000 gigabytes this year. Much of this new data consists of unstructured, rich media files that are difficult to catalogue, archive, develop policies for or place in on-demand access queues for users. Businesses have struggled under the load, throwing more network-attached storage (NAS) resources onto their networks, but it’s getting increasingly more difficult to manage all of these resources internally and to concurrently deliver quality of service (QoS) to requesting users.
The situation is prompting renewed discussions about acquiring on-demand storage networking from outside organizations during times of data overflows, since, by doing so, IT only subscribes to what is needed, is not necessarily committed for long term to a particular vendor and also has the opportunity to fund internal storage network expansion when budgets allow it.
But there is still confusion between what the industry calls “grid” storage networking versus “cloud” storage networking.
In traditional grid storage networks, servers, networks and storage are managed as single or unified entities, with the ability to scale and support high-capacity and even high-performance computing and storage environments. While organizations can grow and interconnect their own internal storage resources to help meet the challenge of heavy storage usage, grid’s integrated systems management framework is able to span locations, systems and software architectures. This simplifies access to both new and archived data. Traditionally, grid storage has been most productively employed in high-performance computing (HPC) environments characterized by educational and scientific research institutions, which perform complex numerical modeling and computations that generate enormous files that must be shared or accessed in an on-demand manner and under strict data usage policies. These files are processed by the grid-in-batch mode and are then made available as soon as possible.
Day-to-day storage usage for normal business enterprises is a little different from this model. While many corporate files might be large because they consist of rich media, some files might also contain transactional data. Most enterprise data files do not begin to approach the size of scientific research files, nor do they require the rigid usage policies. Day-to-day enterprise storage demands also fluctuate. There might be periods of peak demand on the corporate data center where it makes sense to direct overflow data loads to a cloud-based storage network and other days where the demand can be fully met and managed internally. For these situations, cloud-based storage networking works the best because it can be allocated then shut down as needed in real time, and it allows for wider variations in file sizes.
Commercially, major vendors are making their investments into cloud storage infrastructures that focus on virtual storage networks capable of supporting the Internet, high-performance computing and a wide variety of enterprise storage needs. Like grid-based storage, these cloud solutions involve the creation of storage objects and an accompanying metadata database capable of referencing these objects to retrieve the requested data. By incorporating many elements of grid storage, these storage cloud build-outs also very likely will signal a point when both grid- and cloud-based storage solutions coalesce into a single cloud-based platform. This is important for IT because such a convergence will also signal an end to the spiral of complications that invariably arise when companies are forced to sort through different terminologies, when all they want to do is extend their storage networking to external enterprise resources as needed.
Most importantly, commercially available storage networking, likely in a cloud, will become more flexible in the solutions it offers enterprises. Server blades, disk drives and networks in the cloud will continue to work with heterogeneous computing platforms, whether they are Wintel, Linux®, Unix® or even proprietary. Cloud storage network providers will also be borrowing pages from grid computing’s playbook, with more policy-setting options for files and data that can meet enterprise compliance, governance and security needs.
For more information, see:
Stop Overprovisioning with Storage Resource Management
SSD in the Enterprise: The New Face of Tiered Storage
Federated Storage Solutions — What You Need to Know
        | 
 |
